[Legacy Report] IEEE CIS ACT Seminar
Transferring learning of knowledge and functionality from simple problems to larger scale and related problems
Abstract: Transfer learning is an increasingly important concept as the scalability of existing artificial intelligence methodologies is stagnating due to the complexity of modern problems. Individual problems can be solved, but an alternative to many individual computers working in ensembles is required for convenience, cost and elegance. Sub-symbolic approaches can now share models of learned patterns, but not functionality. Symbolic approaches can learn building blocks of knowledge as well as select and construct problem features, but are often restarted for each new problem. The symbolic evolutionary computation technique of Learning Classifier Systems (LCSs) was conceived 40 years ago as a cognitive system, although it is since been developed into a powerful and flexible classification approach. The work presented shows how LCSs can utilise code fragments (building blocks of knowledge) in ‘if-then’ rules to transfer learnt knowledge from small to large scale problems in the same domain. Furthermore, the use of these rules enables functionality learned in sub-problems to be transferred to related problems. Provided the human experimenter can set a rough curricula for learning threshold concepts, the underlying patterns/models in a problem domain can be learnt in an interpretable manner. Solutions to previously intractable problems and general solutions to domains, such as even/odd parity, carry and multiplexer problems, are presented for the first time.
Biography: A/Prof. Will Browne received a BEng Mechanical Engineering, Honours degree from the University of Bath, UK in 1993, MSc in Energy (1994) and EngD (Engineering Doctorate scheme, 1999) University of Wales, Cardiff. After eight years lecturing in the Department of Cybernetics, University Reading, UK, he was appointed to School of Engineering and Computer Science, Victoria University of Wellington, NZ in 2008. Associate Professor Browne's main area of research is Applied Cognitive Systems. This includes Learning Classifier Systems, Cognitive Robotics, and Modern Heuristics for industrial application. Transferring learning of knowledge and functionality from simple problems to larger scale and related problems Abstract: Transfer learning is an increasingly important concept as the scalability of existing artificial intelligence methodologies is stagnating due to the complexity of modern problems. Individual problems can be solved, but an alternative to many individual computers working in ensembles is required for convenience, cost and elegance. Sub-symbolic approaches can now share models of learned patterns, but not functionality. Symbolic approaches can learn building blocks of knowledge as well as select and construct problem features, but are often restarted for each new problem. The symbolic evolutionary computation technique of Learning Classifier Systems (LCSs) was conceived 40 years ago as a cognitive system, although it is since been developed into a powerful and flexible classification approach. The work presented shows how LCSs can utilise code fragments (building blocks of knowledge) in ‘if-then’ rules to transfer learnt knowledge from small to large scale problems in the same domain. Furthermore, the use of these rules enables functionality learned in sub-problems to be transferred to related problems. Provided the human experimenter can set a rough curricula for learning threshold concepts, the underlying patterns/models in a problem domain can be learnt in an interpretable manner. Solutions to previously intractable problems and general solutions to domains, such as even/odd parity, carry and multiplexer problems, are presented for the first time. IEEE CIS Seminar Associate Professor Will Browne Victoria University of Wellington, New Zealand Venue: LT12, Building 32, UNSW Canberra Time: 3:00 - 4:00 pm, March 17, 2016 Register (free) via this link : http://goo.gl/forms/cscUho33Ex School of Engineering and Information Technology Blue skies research includes analogues of emotions, abstraction, memories, Small-Worlds phenomenon, dissonance and machine consciousness. He is an Associate Editor for Neural Computing and Applications, and Applied Soft Computing. He has published over 100 academic papers, including two best paper awards in Genetic and Evolutionary Computation Conference and in IEEE Transactions on Evolutionary Computation on scalable learning.
Date and Time
Location
Hosts
Registration
-
Add Event to Calendar
- Canberra, Australian Capital Territory
- Australia